Soitec: The AI-Photonics Turn Is Real, But the Price Already Assumes It Wins

Fundamentally, Soitec now sits between categories. The core businesses proved highly cyclical, and the earnings compression too severe, for it to keep the clean “high-quality growth” label it once carried. Yet it is not a classic distressed turnaround either: liquidity is solid and the company still owns scarce technology. The most accurate portrait is a company in transition, with credible cyclical-reversal features and an increasingly speculative AI re-rating layered on top. The capital market is no longer pricing Soitec on what it is earning; it is pricing Soitec on what its photonics and AI-adjacent substrate franchise might become if hyperscaler and network-optics capex remain strong.

That leads to the practical judgment. The business itself is more resilient than the fiscal 2026 loss suggests: cash flow, liquidity, and customer relationships all held. The stock is less forgiving than the trough earnings suggest, because the AI story has already pulled a large chunk of future hope into the present price. Soitec is not broken. Whether the present price still leaves a comfortable margin of safety for a balanced investor is a much harder question, and the honest answer is mostly negative.

Vertical history and business model

Origins and listing path

Soitec was born in Grenoble in 1992 out of CEA research. Founders André-Jacques Auberton-Hervé and Jean-Michel Lamure built the company around Smart Cut, a layer-transfer process that turned silicon-on-insulator from an engineering idea into an industrial product. The original problem was not “more wafers” but “better wafers”: chipmakers needed materials that could improve electrical isolation and performance as conventional silicon scaling became harder. A 1997 alliance with Japan’s Shin-Etsu Handotai helped validate the industrial path, and in 1999 Soitec listed on Paris’s Nouveau Marché, later Euronext Paris, while inaugurating Bernin 1, then described by the company as the world’s largest SOI production site.

The early capital-markets story was therefore classic French deep-tech: proprietary process know-how, a research-origin moat, and a category that was small but strategic. The current business model still rests on that foundation, but it is now broader. Soitec no longer sells only classic SOI wafers. It has built a portfolio of engineered substrates for multiple end markets. It has also built an ecosystem position: customer qualification, process integration, and long product cycles now matter as much as physical production. That combination is why the company can still earn premium economics when adoption is rising.

Stages of development

The first stage ran from founding through the mid-2000s. The task was industrial proof. Soitec had to persuade a conservative semiconductor industry that engineered substrates were not laboratory curiosities. Strategic alliances, pilot lines, and Bernin capacity build-out were the tools. The lasting result was a company whose credibility came from process know-how and customer qualification rather than from scale alone.

The second stage, roughly the late 2000s through 2015, was a strategic detour. The company expanded into solar, lighting and equipment markets. Soitec later described this period plainly: it had ventured outside electronics and then chose to refocus on its core business. That wording matters because it captures what the cycle taught management and shareholders. Soitec is strongest when it stays close to engineered-substrate bottlenecks where its IP and process integration matter; it is weaker when it tries to become a broad technology conglomerate. The 2016 capital increase plan of €130 million to €180 million, supported by core shareholders including CEA Investissement, NSIG and Bpifrance, was part of repairing the balance sheet after that failed expansion.

The third stage, from 2016 through 2020, was the repair period. The company refocused on electronics, simplified the portfolio, and re-entered capital markets on better terms. The 2018 €150 million OCEANE financing was a reminder that even high-IP substrate businesses need patient funding when capacity and R&D have to be maintained ahead of demand. Yet this stage also created the platform for the later boom: Singapore expansion, renewed readiness for 300mm growth, and a clearer link between product roadmaps and end-market adoption.

The fourth stage, from 2021 through 2024, looked like the business finally graduating from niche success to broad structural growth. Fiscal 2022 revenue reached €863 million, up 48% from fiscal 2021, with EBITDA margin of 35.8%. Fiscal 2023 revenue rose again to €1.089 billion with a 36.0% EBITDA margin, and fiscal 2024 still managed €978 million with a 34% EBITDA margin despite smartphone inventory correction. Management used that period to invest hard in capacity, expand in Singapore, and add silicon-carbide capabilities through the 2021 NOVASiC acquisition, aiming to open another long growth vector through power semiconductors. Investors began to price Soitec as a high-quality structural grower rather than as a cyclical wafer specialist.

The fifth stage, fiscal 2025 through today, is the painful reset. Fiscal 2025 still looked manageable on the surface: revenue fell to €891 million, but Edge & Cloud AI grew to €216 million and free cash flow remained positive at €26 million. Then the downturn deepened. Automotive and industrial customers kept destocking, RF-SOI inventories stayed too high, SmartSiC ramp timing slipped, and in May 2025 management withdrew full-year and medium-term guidance. Fiscal 2026 brought the full force of the reset: revenue fell to €591.8 million, EBITDA margin slid to 25.4%, current operating income turned slightly negative, and the company booked a large impairment and a deep bottom-line loss. The share-price narrative changed accordingly. This was no longer a premium compounder. It became first a cyclical casualty, then a possible AI-adjacent recovery vehicle.

Financial vertical review

A concise five-year view shows both the depth of the cycle and the reason Soitec is still investable as a business, if not automatically as a stock.

This table is drawn from Soitec’s full-year releases and annual disclosures.

The business reason behind the numbers is straightforward. Revenue growth from FY22 to FY23 was broad-based and carried high operating leverage because the installed base and R&D platform were already there; margins expanded as utilization and mix improved. FY24 and FY25 were not yet collapses, but they showed the vulnerability of Soitec’s “quality growth” image: inventories in smartphone RF and weakness in automotive immediately translated into slower growth, higher working-capital needs and weaker free cash flow. FY26 was the inflection from disappointment to trough. EBITDA almost halved, but operating cash flow held at €202 million because receivables fell by €145 million and inventories fell by €24 million, while capex was cut to €135 million from €230 million. That is a resilience point, but it is not a clean sign of restored earning power. It is evidence that management protected liquidity aggressively while the income statement deteriorated.

Business model, cost structure and moat

Soitec reports by end market, which is the best way to understand the business machine. In fiscal 2025, mobile communications was €546 million, or 61% of sales; automotive and industrial was €129 million, or 15%; and Edge & Cloud AI was €216 million, or 24%. In H1 fiscal 2026, the mix changed sharply: mobile was €119 million, automotive and industrial only €15 million, and Edge & Cloud AI €96 million. That does not mean AI has become the dominant business. It means the legacy businesses shrank so much that AI became relatively more important.

The cost structure is the central analytical point. Soitec has a heavy fixed-cost base: fabs, process engineering, qualification teams, and a large R&D burden. That creates attractive operating leverage when volumes rise, but it punishes profit quickly when utilization falls. Management said exactly that in its FY26 presentation: revenue was down 30% organically, lower fab loading weighed on margins, and the year became a story of execution and cash. The EBITDA margin fell from 33.5% in FY25 to 25.4% in FY26, and current operating income slipped below zero.

The real moat has three parts. First is process IP and accumulated know-how. Smart Cut and related substrate engineering capabilities sit inside years of customer qualification and manufacturing experience, not just patents on paper. Second is ecosystem embedding. Soitec wins when a substrate becomes part of a customer’s process stack and a chip architecture’s bill of materials, because switching costs then include redesign, re-qualification and production risk. Third is the ability to serve small-but-critical substrate niches before they become large enough for a broader field of rivals to care. That is how RF-SOI became a standard in mobile RF and why Photonics-SOI matters now. The weaker “marketing moat” claims are around category breadth. Soitec’s portfolio is wider than before, but its moat does not come from being in many substrates. It comes from being uniquely useful in a few of them.

Governance is mixed but not alarming. The ownership base includes state-linked and strategic industrial shareholders, and the board has representatives from Bpifrance and CEA Investissement. That can support long-term investment and financing access, though it also means governance should not be read through an Anglo-American pure free-float lens. As of March 31, 2026, the shareholding structure included free float at 60.95% of capital, CEA Investissement at 7.19%, Bpifrance Participations at 6.12%, NSIG Sunrise at 5.83%, and The Capital Group at 11.45%, with different voting-right percentages due to double-vote rights. There is no evident accounting scandal in the recent record, but management credibility has clearly been tested by the abandoned FY26 medium-term targets that had looked ambitious only a few years earlier.

Industry and horizontal landscape

Industry structure and cycle

Soitec belongs to several cycles at once, and that is what makes it easy to misread. The mobile RF-SOI business follows a semiconductor inventory cycle tied to smartphones, foundry purchases and channel digestion. Automotive and industrial substrate demand follows a slower capex-and-inventory cycle, and recent evidence from both Soitec and peers shows it has stayed depressed for longer than expected. Photonics-SOI and AI-related optical interconnect demand follow a different rhythm: hyperscaler capex, network upgrades, and optical packaging adoption. Those are not the same cycle, not the same customer base, and not the same margin profile.

The broad silicon-wafer industry backdrop remains uneven. SUMCO said in February and May 2026 that 300mm leading-edge demand tied to AI was improving, but legacy demand and 200mm conditions remained weak, with customers still drawing down inventories. Siltronic’s FY2025 results showed sales down 4.7% year on year, negative EBIT, and only a Q4 volume rebound helped by timing shifts, pricing pressure and product mix. Shin-Etsu’s semiconductor materials segment, by contrast, showed better resilience because of portfolio breadth and scale. This peer evidence matters because it says the problem is not unique to Soitec. The cycle is real. What is unique is Soitec’s much stronger exposure to specific substrate architectures and its much more dramatic AI optionality.

Geopolitics cuts both ways. China is simultaneously an opportunity and a risk. Opportunity, because Soitec extended its licensing framework with NSIG for another 10 years in March 2026 in response to Chinese demand for SOI products, and in June 2026 signed a new 300mm BCD-on-SOI collaboration with ZenSemi targeting AI data centers, EVs, robots and industrial applications. Risk, because strategic substrate IP, Chinese localization policy and cross-border technology restrictions could all complicate that same opportunity. The NSIG extension explicitly stated there was no new technology transfer in outside reporting, which underlines how sensitive this area has become.

Horizontal competitor portrait

Soitec has direct peers, but only partly. Shin-Etsu, SUMCO, GlobalWafers and Siltronic are the right cross-checks for cycle discipline, customer behavior and wafer-industry valuation. None is a true one-for-one substitute, because Soitec’s differentiation sits in engineered substrates rather than broad commodity silicon-wafer supply. The right horizontal conclusion is therefore not “who is bigger,” but “what each company became.”

Shin-Etsu became the scale-and-reliability incumbent. Customers pick it because it can supply critical semiconductor materials at enormous scale across multiple categories, not because it owns a niche substrate story. That breadth stabilizes results and gives it more room to absorb downturns. Soitec cannot match that breadth, but it does not need to when the substrate itself changes device economics.

SUMCO stayed the focused silicon-wafer pure play, chosen for high-end silicon wafer capability and long-standing relationships in leading-edge manufacturing, especially for 300mm. Its own commentary makes clear that AI is helping leading-edge 300mm, but legacy and smaller-diameter products remain weak. That is the closest cycle analogue to Soitec’s current situation: one part of the book already turning, another still in correction. The difference is that SUMCO is still a silicon-wafer specialist, while Soitec is monetizing architecture-specific substrate advantages.

GlobalWafers built its position as the geographically diversified expansionist, valued for footprint, long-term agreements and capacity optionality, especially with its U.S. expansion. Management said in Q1 2026 that 12-inch demand remained strong and visibility there was solid, with advanced nodes and AI supporting utilization. That is a reminder that wafer markets with real AI pull are already rewarding capacity in certain formats. But GlobalWafers still competes on manufacturing breadth and location more than on a substrate architecture moat.

Siltronic, for its part, is the disciplined European volume player: customers stay for quality, long-term contracts and execution in silicon wafers, not for a disruptive product roadmap. Its recent results show how ugly the cycle can still look even for a much more conventional wafer supplier: revenue pressure, negative EBIT, and heavy depreciation from new Singapore assets. If anything, Siltronic is a warning that a cyclical rebound can lift volumes before it restores earnings quality. That warning also applies to Soitec.

A compact numbers snapshot helps, but the prose carries the real point.

Sources for this table are current market pages and each company’s cited releases.

Soitec sits between specialty-IP tech and materials cyclicality, and that in-between position is what explains the valuation gap with peers. Peers are easier to underwrite because their products are closer to “must-have production input.” Soitec is harder to underwrite because its upside is larger if a substrate standard wins, but the downside is harsher when an adoption timetable slips or an end-market pauses. That is why the market can give Soitec a premium in good times and then cut it far faster in bad times. Today’s price says investors are once again granting Soitec a meaningful structural premium. The unresolved question is whether the earnings base has caught up with that premium.

Current fundamentals and valuation

What is actually happening now

The last four quarters show a business that improved sequentially all year but did not yet heal.

This sequence comes from company releases and Reuters’ summary of Q4.

The numbers read like a bottoming process, but not yet a clean recovery. In Q1, mobile was only €43 million, automotive and industrial €5 million, and Edge & Cloud AI €44 million. In Q2, mobile rose to €76 million and AI to €52 million, while automotive and industrial remained just €11 million. The pattern says Soitec is climbing out of an unusually low base, with AI helping at the margin and traditional businesses still far from normal. It does not yet prove that demand has normalized.

The market is trading the stock on a narrower set of facts. It is trading the quarterly revenue staircase, the photonics signal, the new CEO’s cash-discipline message, and the symbolic importance of positive free cash flow during a trough. That is real fundamental progress, but it is not the same thing as broad profit recovery. Operating cash flow in FY26 stayed flat at €202 million while EBITDA fell from €298 million to €151 million, because a large working-capital release offset the earnings decline and capex was cut from €230 million to €135 million. That improvement deserves credit, but it should not be mistaken for a return to steady-state profitability.

Estimate revisions and target prices reflect that tension. After the 2025 guidance withdrawal, Jefferies cut Soitec to hold and lowered its target to €50 from €58, citing uncertainty in cyclical recovery. By July 2026, however, public aggregation sites showed a much higher and widely dispersed analyst target set, with Investing.com listing a 19-analyst average target of €141.79 and a neutral consensus, while other aggregators showed broad dispersion between low and high targets. The right conclusion is not that consensus is “right,” but that the stock has become a battleground name whose valuation depends heavily on how much photonics and AI revenue investors are willing to capitalize today.

Cash-flow passthrough and historical valuation context

Over the last five fiscal years, Soitec’s operating cash flow has been more stable than net income, but not because the business suddenly stopped being capital intensive. FY22 operating cash flow was €255 million, FY23 €263 million, FY24 €166 million, FY25 €202 million, and FY26 €202 million. Free cash flow over the same period was €42 million, €34 million, negative €43 million, €26 million and €63 million. The key point is that net income and owner earnings have diverged sharply at different parts of the cycle because capex, leasing-financed tools, working capital and occasional one-offs move around a lot.

Soitec does not disclose a maintenance-versus-growth capex split, so the owner-earnings bridge must be inferential. A reasonable working assumption is that around €80 million to €100 million of annual capex is maintenance or refresh spending for a business with multiple fabs, process migration, and ongoing substrate qualification work, while capex above that is more growth-oriented. On FY26 operating cash flow of €202 million, that implies owner earnings of roughly €98 million to €118 million before any normalization of working capital. Against a current market cap of about €4.20 billion, that is roughly a 2.3% to 2.8% owner-earnings yield. France’s 10-year government bond yield was around 3.7% in early July 2026. That comparison is not flattering: at the current price, investors are accepting an owner-earnings yield below the sovereign benchmark and are relying on future growth to justify it.

On trailing market metrics, the stock is plainly expensive for a trough-multiple investor. At €117.30 per share and roughly €56 million of net debt, Soitec trades near 7.2x trailing EV/sales and around 28x trailing EV/EBITDA on FY26 numbers. Those are not distressed multiples. They are recovery-and-optionality multiples. That does not make them irrational; it means the market is already looking through the trough.

Peer valuation and absolute valuation

For a loss-making or near-loss-making cyclical semiconductor-materials name, trailing P/E is unusable. The right toolkit here is trailing EV/sales, normalized EV/EBITDA, and a cross-check against owner earnings. The primary method below uses EV/sales because Soitec’s current market narrative is driven more by future mix than by present margins, while the secondary check uses normalized EBITDA.

This is valuation-scenario analysis within a research framework, not investment advice. It uses current equity value inputs from July 2026 and Soitec’s FY26 financial base.

Put plainly, Soitec deserves a premium to conventional wafer peers when its substrates are creating or defending architecture-level profit pools. But the current price already assumes either a meaningful cyclical rebound or a larger photonics-led strategic shift. The base case does not leave much room for mistakes. The optimistic case can justify numbers above today’s price, but only if photonics and AI-adjacent products become a much larger share of group economics without a fresh capex or pricing problem.

Expectation gap and margin-of-safety check

The expectation embedded in the stock now is not “FY27 will still be messy.” The expectation is “FY27 is a bridge year to a much better mix and much better earnings power.” That is why the next earnings print matters less for the absolute revenue number than for three details: Edge & Cloud AI growth, evidence that RF-SOI inventories are genuinely clearing, and any sign that automotive demand is moving from base-effect improvement to real order recovery.

The most fragile assumption in the base case is the size and durability of the photonics ramp, not mobile stabilization. If that assumption is cut to 70% of what the market currently seems to hope for, the base-case fair value drifts much closer to the low end of the €80–105 range and can easily fall below €90. That is because the premium multiple depends on AI-photonics eventually being large enough to change the company’s average quality, not merely provide a useful side business.

On the explicit margin-of-safety tests: the current price is well above the value implied by the conservative scenario, so the margin of safety against that scenario is zero. If owner earnings stay flat around the €100 million area for the next three years, the implied earnings yield at today’s equity value remains below the French 10-year government bond yield around 3.7%, which means there is no margin of safety at this buy price. This is therefore a good-company-but-bad-price setup more than a broken-business setup. Margin-of-safety sufficiency verdict: none.

Risks, catalysts, and tracking dashboard

Permanent-loss risks that matter

The first real risk is that the RF-SOI correction reflects a structural downgrade in Soitec’s earnings power, not just cyclical inventory digestion. Probability: medium. Impact: high. Observable indicators: mobile communications revenue failing to rebuild above roughly €150 million per half-year, explicit price/mix deterioration in RF-SOI, or customer concentration becoming more visible through repeated shipment pauses. Transmission path: lower volume and weaker price/mix would cut fab loading, keep EBITDA margins depressed, and compress the premium multiple because the market would stop treating mobile as a recoverable profit pool.

The second is that automotive and industrial take much longer to recover than management and investors want. Probability: high. Impact: medium to high. Observable indicators: Automotive & Industrial revenue remaining stuck near Q1–Q2 FY26 levels, no meaningful Power-SOI 300mm ramp progress, and continued commentary about inventories or qualification delays in SmartSiC. The transmission path is slower than in mobile, but just as damaging. These products were supposed to diversify the business; if they stay dormant, Soitec remains too dependent on mobile and too narrow an AI beneficiary.

The third is hype risk in photonics and AI. Probability: medium. Impact: high. Observable indicators: Edge & Cloud AI growth slowing abruptly, customer concentration inside photonics rising, or management talking more about “ecosystem intimacy” than shipped revenue. The stock has already rerated as if photonics will become central. If actual revenue keeps rising but not fast enough to re-weight company economics, valuation can compress even while the business improves modestly.

The fourth is cash-flow reversal after the easy balance-sheet clean-up. Probability: medium. Impact: medium to high. Observable indicators: operating cash flow falling below EBITDA, inventories rebuilding, receivables rising again, or capex moving back above about €180 million without clear incremental revenue proof. FY26 free cash flow held up partly because management squeezed working capital and cut capex. If the business later needs another investment wave before revenues recover sufficiently, the market may discover that trough cash generation was more tactical than structural.

The fifth is geopolitical and IP risk around China. Probability: medium. Impact: medium. Observable indicators: delays or softening in NSIG-related execution, tighter export-control rhetoric around advanced substrate technologies, or shifts in Chinese localization policy. Soitec’s Chinese partnerships can accelerate growth, but they also increase the sensitivity of its IP and licensing model to policy. A technology supplier can lose valuation support quickly if investors begin to fear that regional growth comes with weaker control over the long-term profit pool.

Catalysts and tracking dashboard

Positive catalysts are clear enough: a clean Q1 FY27 print on 2026-07-28 showing continued sequential growth, evidence that Edge & Cloud AI is sustaining double-digit organic growth excluding Imager-SOI, a meaningful mobile restock without negative price/mix, or any hard commercial proof that photonics wins are broadening beyond a few programs. Negative catalysts are equally clear: a stall in sequential revenue growth, renewed guidance caution, AI revenue that grows but disappoints versus stretched expectations, or a fresh rise in capex before margin recovery has arrived. The next scheduled earnings-related event is Q1 FY27 revenue on July 28, 2026.

The point of this dashboard is to separate real improvement from narrative reinforcement. Revenue alone is not enough; the mix must improve. AI growth alone is not enough, either; it has to become large decisively faster than the legacy books recover. And free cash flow only counts if it stays positive after working-capital normalization. That is the discipline required for a stock that has already rerated on possibility.

Cross-synthesis, research uncertainties, sources, and other tickers

Cross-synthesis summary

Vertically, Soitec has proven one thing beyond doubt: it can create and industrialize engineered substrates that matter. That capability survived a failed diversification into solar and lighting, a balance-sheet repair, a new round of capacity investment, and now a severe cyclical downturn. The company’s deepest competence is turning specialized materials science into manufacturable substrates that can become embedded in customer process flows for years, not forecasting demand. That is a genuine capability and the reason the business still deserves close attention after such a brutal fiscal 2026.

Its past success, though, came from more than technology. It came from a favorable combination of tailwinds: 5G mobile RF complexity, automotive semiconductor content expansion, and investor willingness to pay premium multiples for any semiconductor bottleneck with visible growth. What fiscal 2026 proved is that not all of those tailwinds were structural. Mobile RF inventories can overshoot. Automotive programs can pause longer than planned. New substrate ramps such as SmartSiC can slip. When that happens, Soitec’s fixed-cost structure and premium valuation both work in reverse.

Horizontally, the company’s real advantage versus conventional wafer peers is technological specificity, not scale. Customers do not buy Soitec because it is the largest substrate supplier in the world. They buy it when a specific device architecture performs better on its substrates. That is powerful when adoption is expanding. It is also why the weakness can be structural if an architecture loses momentum. The current evidence is mixed, not fatal. RF-SOI and automotive are still under pressure, but photonics and AI-linked FD-SOI demand are now big enough to matter. The weakness therefore looks temporary in liquidity terms and cyclical in end-market terms, but still structurally important in valuation terms because it revealed how narrow the company’s old profit engine really was.

The market is most likely misjudging pace, not direction. It is probably right that Soitec has more strategic value in AI optical infrastructure than the market appreciated in 2025. It is probably too optimistic about how quickly that value can dominate the income statement. The next year will be about whether sequential improvement can continue without a broad cyclical recovery. The next three years will be about whether photonics and Power-SOI become large enough to change company quality. The next five years will decide whether Soitec is primarily an AI-era enabler with a diversified substrate profit pool, or simply a brilliant niche supplier whose earnings remain too cyclical for the premium it often commands.

Bull and bear reasons

Bull reasons:

• Soitec exited its trough year with €562 million of cash, only €56 million of net debt and €120 million of undrawn credit lines, which gives it time to invest through the cycle.
• Edge & Cloud AI is growing for real: excluding the Imager-SOI phase-out, that division grew 34% organically in H1 FY26, and Q4 commentary again pointed to photonics strength.
• Fiscal 2026 revenue improved sequentially every quarter from €92 million to €202 million, consistent with a trough-to-rebuild pattern rather than a business in free fall.
• Soitec’s moat sits in architecture-specific substrates, which can preserve pricing power and stickiness better than broad commodity wafer supply when adoption keeps advancing.

Bear reasons:

• The old core is still damaged: H1 FY26 mobile revenue was only €119 million and automotive and industrial just €15 million, with management still citing inventory correction and weak demand.
• FY26 free cash flow looked healthy partly because receivables fell, inventories were cut and capex was reduced, while EBITDA itself fell by almost half.
• The current valuation already discounts a material recovery, with trailing EV/sales around 7.2x and owner-earnings yield below the French 10-year bond yield.
• Leadership has changed, but the new CEO is still too early in his tenure to have earned full execution credibility on a cycle turn.

Pre-mortem

A three-year, down-50% script is easy to write because the ingredients are already visible. First script: by 2027, RF-SOI inventory correction ends only nominally, but mobile revenue fails to recover because customers stay cautious and pricing weakens. Automotive remains sluggish, SmartSiC stays delayed, and photonics grows but from too small a base to offset the hole. Revenue stalls in the €650 million to €700 million area, EBITDA margin stays around 22% to 24%, and the market stops paying AI-premium multiples, compressing EV/sales from about 7x toward 4x. The share price could fall into the €55 to €70 zone.

Second script: hyperscaler optical capex cools after a burst of AI spending, so Photonics-SOI growth slows sharply in 2027 or 2028 just as investors are treating it like Soitec’s new core. At the same time, mobile and automotive improve only gradually. The company then looks like a still-cyclical materials supplier with better stories than earnings. The multiple compresses before earnings normalize, and a stock that had been priced for transition to “AI infrastructure enabler” is repriced back toward “specialty wafer cyclical.”

Final research conclusion

Soitec today is a rare-substrate company with a real technology franchise and a temporarily damaged earnings base. The market is right to reject the lazy view that fiscal 2026 proves structural decline. Cash generation, liquidity, customer programs and photonics traction all say the company still matters. The harder question is the stock, not the business. At the current price, investors are paying as though the photonics and AI vector will do more than cushion the trough. They are paying as though it will change the company’s quality fast enough to justify a premium multiple before the legacy businesses have fully recovered. That is possible. It is not yet cheap.

My main concern is timing and valuation, not solvency or product relevance. The next twelve months can still deliver good operating news without creating good shareholder returns if the current price already capitalizes most of that news. What would change my mind? A materially lower entry price, or a few more quarters showing that Edge & Cloud AI can scale fast enough to reshape group economics while mobile and automotive stop destroying utilization. Until then, the right posture is interested, not aggressive.

【Company-profile scores】

• Fundamental quality: medium
• Growth: medium
• Moat: medium
• Financial soundness: medium
• Management credibility: medium
• Valuation attractiveness: low
• Risk level: high
• Suitable investor type: cyclical

【Investment rating】

• Rating: Watch
• One-line thesis: Strong substrate IP and real AI-photonics traction are offset by unresolved core-market weakness and a share price that already discounts substantial recovery.
• 【Ideal Buy Price】60–75 EUR Basis: roughly 20% or better below the value implied by the conservative scenario, where recovery is partial, mobile and automotive remain subdued, and the market pays only a restrained specialty-materials multiple.
• Acceptable hold price: 80–105 EUR
• Clearly overvalued price: 132–145 EUR
• Current-price classification: outside the three bands
• Whether to wait for a better price: yes. A buy would require either a pullback into the 60–75 EUR zone, or proof that Edge & Cloud AI can scale enough to lift the base-case value range materially. The opportunity cost of waiting is missing a continued AI-driven rerating; the benefit is avoiding a zero-margin-of-safety entry.
• Target holding horizon: 3–5 years
• Expected annualized return: conservative about -17%; base about -8%; optimistic about +6%, using midpoints of the scenario ranges over a three-year horizon and excluding dividends.
• Max-loss risk: about 50%–55%, triggered by a failure of mobile and automotive recovery combined with a slowdown in photonics-led AI demand that compresses valuation toward 4x sales.
• Reassessment-trigger signals: if Edge & Cloud AI stops growing organically; if Mobile Communications stays below about €70 million a quarter for two consecutive quarters after Q4 FY26; if Automotive & Industrial does not improve meaningfully by early FY27; if EBITDA margin stays around the mid-20s despite rising revenue; if capex re-accelerates above about €180 million without visible demand support.

【Valuation Range】

• current: 117.30 (close as of 2026-07-03)
• bear (conservative · ideal buy zone): [60, 75]
• base (fair · acceptable hold zone): [80, 105]
• bull (optimistic · above the clearly-overvalued line): [132, 145]

Research uncertainties

The main blind spots are practical rather than conceptual. First, the exact IPO pricing and valuation from the 1999 listing were not recovered from an easily accessible primary document in this session, though the listing path itself is clear from company history materials. Second, Soitec’s FY26 mix is easier to reconstruct quarter by quarter than through one clean full-year end-market bridge, because different releases emphasize H1, Q4 and product-level commentary rather than one single annual mix slide in parsed text. Third, third-party consensus target data are unusually dispersed and change quickly because the stock’s 2026 rerating has been extreme; that makes any consensus number less stable than usual. Fourth, the maintenance-versus-growth capex split is not disclosed by management, so owner-earnings analysis necessarily uses an informed estimate rather than a reported figure.

Sources

Primary sources used most heavily were Soitec’s FY22, FY23, FY24, FY25 and FY26 annual result releases, the 2025-2026 half-year report and press release, the FY26 presentation, the FY26 Universal Registration Document, the investor-relations calendar and voting-rights disclosures, the CEO succession and CFO appointment releases, and Soitec’s corporate history and shareholding pages.

Secondary and market-context sources used were Reuters, The Wall Street Journal, Google Finance, Investing.com and MarketWatch pages for current prices, market caps, bond-yield context, peer-market data and contemporaneous reporting on Soitec’s share-price moves and peer operating conditions. Peer company materials used included Shin-Etsu, SUMCO, GlobalWafers and Siltronic investor documents and result releases.

Other tickers mentioned

• 4063.TSE: Shin-Etsu Chemical, used as the scale-and-reliability incumbent in semiconductor materials.
• 3436.TSE: SUMCO, used as the closest cycle analogue among focused silicon-wafer suppliers.
• 6488.TW: GlobalWafers, used as the geographically diversified wafer peer with U.S. expansion optionality.
• WAF.DE: Siltronic, used as the European wafer peer showing how slow volume recovery can still coexist with weak earnings.
• GFS.US: GlobalFoundries, mentioned because Soitec supplies RF-SOI into GF’s 9SW radio platform.
• SWKS.US: Skyworks, mentioned because Soitec secured a multi-year POI wafer supply agreement for Sky5.
• STM.PA: STMicroelectronics, mentioned through SmartSiC cooperation and European substrate ecosystem context.
• IFX.DE: Infineon, mentioned because Laurent Rémont joined Soitec from Infineon before becoming CEO.

This report is based on public information and does not constitute investment advice. Markets carry risk; invest with caution.